how to create a grammar containing the text to train

I've read the blog of Eric about How to train the SAPI recognizer, and followed the pseudocode. But I don't know how to create a grammar containing the text to train. Now I have the correct transcript of the audio, but I don't know how to connect this correct transcript file to the grammar. Do I need to c开发者_Go百科reate a XML file? Could you tell me the interface name? Thank you so much.

Read the transcript file into a string, and then use ISpGrammarBuilder::AddWordTransition to define a simple grammar containing the string. Then process the audio from the wavefile, and you should get a recognition event.